Professional Documents
Culture Documents
Algebra Can Be Fun PDF
Algebra Can Be Fun PDF
Perelman
CAN
BE
FUN
.' MIHPublishers
� Moscow , ,
•••••• •••••••• •••••
Ya.1. Perelman
ALGEBRA •
CAN !
•
BE •
FUN J
I Edited and supplemented
by V. G. Bolt.yailsky
Translated
from the thirteenth Russian edition
by George Yankovsky
5
Contents
7
Rafts and a Steamboat 70 Guessing a Birthday 126
Two Cans of Coffce 71 Selling Chickens 128
A Question o[ Dancing 72 Two Numbers and Four Opera-
ileconnaissancc at Sea 73 tions 131
At the Cycle Track 75 What Kind of Rectangle? 132
A Competition of Motocyclists 76 Two Two-Digit Numbers 133
A verage Speeds 78 Pythagorean Numbers 135
High-Speed Computing Ma- An Indeterminate Equation of
chines 80 the Third Degreo 139
8
A n Optimal Highway 175 Paying for the Wounds of a
When Is the Product a Maxi Soldier 204
mum? 177
When Is the Sum a Minimum? 181
Cha pte r IX. The Seventh Mathe-
matical Operation 206
A Beam of Maximum Volume 181
Two Plots of Land 182 The Seventh Operation 206
Making a Kite 183 Rivals of Logarithms 207
Building a House 185 The Evolution of Logarithmic
Fencing in a Lot 187 Tables 209
A Trough of Maximum Cross Logarithmic-Table Champions 210
Section 188 Logarithms on the Stage 2H
A Funnel of Maximum Capac- Logarithms on a Stock�naising
ity 190 Farm 213
The Brightest Illumination 192 Logarithms in Music 215
The Stars, Noise and Loga
Astronomical Numbers
Hardly anyone makes as much use of the fifth mathemati
cal operation as astronomers, who are constantly dealing
with enormous numbers that consist of one or two significant
digits and then a long string of zeros. If we were to write out
in full all of these trully "astronomical" numbers, it would
be very inconvenient, especially in calculations. Take the
distance to the Andromeda Nebula. Written out in full in
the ordinary way, we have the following number of kilo
metres;
95 000 000 000 000 000 000.
N ow if we were to express this distance in a smaller unit,
say, centimetres-and this often happens in astronomical
calculations-we would have to add five more zeros:
9 500 000 000 000 000 000 000 000.
• For more about these things see my book Recreational Mechanics
(in Russian),
12
The masses of stars call for still bigger numbers, especially
if they are expressed in grams (and many calculations call
for grams). The mass of tlle sun expressed in grams comes to
1 983 000 000 000 000 000 000 000 000 000 000.
It is easy to imagine the difficulties involved in calcula
ting with such unwieldy numbers and also the cbances of
making mistakes. Yet the above-mentioned numb ers are
by far not the largest to be found in astronomy.
The fifth mathematical operation serves as a simple way
out of this complication. The number one followed by a
string of zeros is always a definite power of ten:
100 = 10" 1000 = 10', 10 000 = 10' and so forth.
The two giant numbers given above can now be written
down neatly as
95 . 102' for the first number and
1983 . 1030 for the second number.
This is done not only to save space but also to simplify
calculations. If we needed to multiply these two numbers
together, all we would have to do is find the product
95 ·1983 = 188 385 and put it in front of the factor
10 23+30 = 10":
95 . 1023 . 1983 .1030 = 188 385 ·10".
This is of course much more convenient than to write out a
number with 23 zeros and then one with 30 zeros and, final
ly, with 53 zeros. Not only is it more convenient but also
more reliable, since it is easy to miss ODe or two zeros in
such long strings of them, and the result would b e wrong.
13
How lUnch does
the Eal'th's Atmosphere Weigh?
Here is a convincing way to simplify practical calculations
by means of exponential notation (using exponents to ex
press powers of numbers): let us determine how many times
the mass of the earth is greater than the mass of the earth's
atmosphere.
As we know, each square centimetre of the earth's sur
face supports a column of air equal to one kilogram . The
atmospheric shell of the earth is, as it were, made up entirely
of such columns of air-as many as there are square centi
metres on the earthIS surface. That is how many kilograms
the atmosphere of our planet weighs. From a reference book
we find the earth's surface to be equal to 510 million, or
51 ·10', square kilometres.
Now figure out how many square centimetres there are in
a square kilometre. A linear kilometre contains 1 000 metres
with 100 centimetres in each, which means it is equal to
10' cm, and a square kilometre contains (10')' � 10lO square
centimetres. And so the earth IS surface works out to
51 ·10' . 1010 � 51 ·10"
square centimetres. And that also is the weight of the earth's
atmosphere in kilograms. Converting to (metric) tons, we
get
51 . 1017: 1000=51 · 10" : 103 = 51 . 10"- 3 = 51 . 10 "'.
N ow the mass of the earth is taken to be
6 . 1021 tons.
To determine OOW much heavier the globe is than its
atmosphere, we perform the following division:
6 · 10": 51 · 10 " "'" 10',
14
which means that the mass of the atmosphere is roughly one
millionth of that of the earth. *
4
=
15
which is about a quarter of a quintillion seconds (quintil
lion as in the American and French system of numeration).
There are 30 million (or 3 . to') seconds in a year, and so
1
A Combination I.ock
PROBLEM
19
The Results 01 Repeated Donbltn!l
A striking example of an exceedingly fast build-up of some
small quantity when repeatedly doubled is the famous
legend about tbe award to be given to the discoverer of
chess. * Here are some other examples, less famous.
PROBLEM
The infusorian paramecium divides in half on the average
every 27 bours. If all newly born infusorians remained alive,
how long would it take for the progeny of one paramecium to
fill up a volume equal to that of tbe sun?
Starting data: the 40th generation of a paramecium, when
none perisb, occupies one cubic metre; we take the volume
of the sun as equal to 10" cubic metres.
SOLUTION
The problem reduces to determining how many times 1 cu
bic .metre has to be doubled in order to obtain a volume of
10" cubic metres. Since 210 "" 1000, we bave
10" � (10')' "" (210)' 290.=
20
Imagine the sun to divide in half, then the halves also
i n half, and so on. How many divisions will it take to reduce
the sun to the size of an infusorian?
Although the reader already knows the answer, 130, it
seems strikingly out of all proportion.
This same problem was proposed to me in the following
form.
A sheet of paper is torn in half, one of the halves is again
torn in half, an� so on. How many divisions will it take to
reduce the paper to the size of an atom?
Assuming the paper weighs one gram and an atom is tak
en to be on the order of 1/10" gram, we reason as follows.
Since it is possible to replace 10" by the approximately
equal expression of 280, it is clear that only 80 halving ope
rations will be required, which is nowhere near the millions
that one often hears as an answer to the problem.
21
curront flows through tho right-hand tube, we say tbat the
circuit is in "state 0", if current flows, then i t is in "state 1" .
To start, let us suppose the circuit is in state 0, which
means the left-hand tube is conducting (Fig. 1). After the
first pulse, current will flow through the right-hand tube
Slarting state a
10 %1
7 sl pulse
% 01
I
Response pulse 2nd pulsr
Fig. I
-1,--O_O---,H °
3rd trigger
°
2nd trigger
H° 1 st
°
trigger
�s
Fig. 2
CQ C!::l �I,-----,
l�j-�r11
4th trigg�r 3rd trigger 2nd trigger
r-
1st trigger
sum
Fig.�3
27
The Number of All Possible
Chess Games
Let us make an approximate calculation of the total num
ber of different chess games that can ever be played on a
chess board. An exact calculation is quite beyond us, but we
will acquaint the reader with a very rough estimate of the
total number of possible chess games. The Belgian mathema
tician M. Kraichik makes the following calculation in his
book entitled The Mathematics of Games and Mathematical
Diversions.
"In the first move, white has a choice of 20 moves (16 moves
for the eight pawns, each of which can move one or two
squares and two moves each of the two knights). For each
move of white, black Can respond with one of the same
20 moves. Combining each move of white with each move
of black, we have 20·20 � 400 different games after the
first move of each side.
"After the first movo, the number of possible new moves
increases. For example, if white made the first move P-K4,
then it has a choice of 29 moves the next time. The number
of possible moves continues to increaso. For instance the
queen alone, standing on Q5 has a choice of 27 moves (on
the assumption that all squares that it can move to are ya
cant). To simplify matters, let us assume the following
average numbers:
20 possible moves for both sides in the first five moves;
30 possible moves for both sides in all subsequent moves.
"Also, we talm the average number of moves in a single
game to be 40. Then we get the following expression lor the
number of possible games:
(20· 20)' , (30 . 30)35".
To get an approximate idea of this figure, let us make a
few simplifying transformations
(20 . 20)'. (30 · 30) " � 20" · 30" � 2"·3" . 1080
.28
Replace 210 b y t he approximation '1000, or 10', an d ex
press 370 as
370 = 3"·;)""" to (3')17 "" '10·80"·= 10.817.'1 0" = 2"· to"
� 2 (210 )'. to" "" 2·10"· to" � 2·10".
And so we obtain
(20·20)'. (30·30)35 "" 10'·2.10".1080 � 2 ·10U6 .
This number leaves far behind the legendary number of
grains of wheat asked as payment for the invention of chess
(that number was a mere 2" - 1 "" 18. 10"). If the whole
population of the world played chess round the clock making
ODe move every second, then it would take no less than
10'00 centuries to exhaust all the games in this marathon
of chess!
L�g in � pawn
King +200 points Pawn +1 poi nt
Quee'n +9 poin ts -0.5 poi nt
Rook -\-5 points Iso ate pawn -0.5 point
Bishop +3 points Doubled pawn - 0 .5 point
K n ig ht +3 poi nts
31
't rue, a machine that can "think through" only three
moves ahead is a rather weak "player". (Good chess masters
think through 10 and morc moves ahead ill their combilla
tions .) But on the other hand, rapid progress is being made
in the computer [ield ami we can expect much better com
puter players in tbe near future.
There are a great many more interesting things in compu
ter-played chess, but they are all far beyond the scope of a
book like this. In the next chapter we will consider some
elementary computer programs.
'l'hree Thl'ees
PROBLEM
PHOBLEM
Write down the largest number using only throe fours
and dispensing with any operational signs.
3 -089 1 33
SOLUTION
If you try the pattern of the two preUlding problems,
you will be wrong because this time the tower of three fours,
4 I' ,
does yield the largest possible number, True enough, 4' =
Let us now figure out for what value of a the last arrange
ment depicts a larger number than the first arrangement.
Since both expressions are powers with equal integral bases,
it follows that the greater quantity corresponds to the
greater exponent. Now when is
a" > l 1a?
34
Divide both sides of the inequality by a to get
aa-{ > 1 1 .
•• 35
Four 'I'wos to l\lake tIle Biggest Numbet·
PROBLEM
Continuing this series of problems, let us now try four
twos. What arrangement of four twos depicts the largest
number?
SOLUTION
Altogether there are eight possible combinations:
2222, 2222, 2222, 2222.
22
222 2 ,
22 22
22 , 22 , 22 .
Which one represents the largest number?
Let us try the top row ftfst, that is, numbers in two
stories.
The ftfst one, 2222, is clearly less than any of the others.
To compare the next two: 222' and 2222 , transform the
second thus:
2222 � 22 " " � (222) " � 484".
The last number exceeds 222' because both the base and
the exponent of 4841 1 are greater than in the case of 222' .
Now compare 2222 with the fourth number of the first
row, 2222. We replace 2222 by a greater number 3222 and then
show that even this greater number falls short of 2'22.
True enough,
3222 � (25)22 �
2 '10
which is a smaller number than 2222 (the exponent is small
er).
And so 'we have the largest number in the top row: 2"' .
That leaves us with five numbers to compare : 2222 and
the following four numbers:
22 2 2 , 22 22 1 2 222 , 2 22
2 .
36
The last number, equal to 216, is too small to continue
the contest. Now the first number is equal to 22' and is
less than 32' or 2". It is less than each of the two numbers
that come after it. Thus we have three numbers for com
parison and each is a power of 2. It is quite evident that the
number with the largest exponent is the largest number.
Now of the three exponents,
222, 484 and 2 20+ 2 ( � 2 1 O · '. 2' ''''' 106 . 4)
the last one is clearly the largest.
And so the largest number that can be depicted by means
of four twos is
2
22 2 .
We can get a rough picture of the size of this number
without resorting to logarithmic tables if we take the fol
lowing approximate equality
210 "" 1000 .
Then
222 � 220 • 2 ' "'" 4 . 10',
In ordinary laITguage:
I In lhc language or algebra:
I
A merchant has a certain sum x
of money.
he then I
iOO pounds.
Duri ng the first year he spent x - l00
I
To the remaining sum x - l00 4x- 400
added one third of it . (x- l00) +
3 3
4x- 700
3
100�
4x-700
3
16x-2800
sum by onc third of it. +
3 9 9
agaiJl
During the third year he
spent 100 pOD nds.
I1 1 00, - 2800
9
100=
16x- 3700
9
38
In ordinary language: In t.he language or algebra;
In ordinary language:
x
of Diophant.us. 'tis a mirach�
that numbers can measure
the length of his life.
In ordInary langufll;(;:
I In the language of algebra:
-x7
I
A sev enth part he spent in
childless wedlock.
SOLUTION
Solving the equation, we fmd that x � 84 and we also
learn that he married at the age of 21, became a father at
the age of 38, lost his son when he was 80 and died when he
was 84.
40
The Horse and the llinle
PROBLEM
Here is another old problem that can readily be translated
from words into algebraic symbols.
"A horse and a mule, both heavily loaded, were going
side by side. The horse complained of its heavy load. 'What
arc you complaining about?' replied the mule. 'If I take one
sack ofl your back, my load will become twice as heavy as
yours. But if you remove one sack from my back, your load
will be the same as mine.'
"Now tell us, wise mathematicians, how many sacks was
the horse carrying and how many sack", the mule?"
SOLUTION
I
If take one sack,
I x-I
m y load
I y+l
your load
I x+l
y-2
the second has two rubles taken away.
I
the third has his sum doubled,
I 2,
t
the fourth ha:\ hi:! sum ha lved,
I "2
I t
then all four brothers will ha ye the same
sum of m oney _
x + 2� y - 2�2'� "2
t � 2x + 4.
42
Then, substituting these values into the first equation, we
obtain
43
Fig. 4
Fig. 5
f1sh appeared at a distance of 20 cubits from the palm tree
which is 30 cubits high.
44
Out for a Stl'oll
PRODLEM
SOLUTION
2 .4
16 - 12 = 4
whence x = km, which is the distance between the
young man's houso and the doctor's house.
Making Hay
A. V. Tsinger, a prominent physicist, rocalls Lev Tolstoy
posing the following problem, one that tho great writer
liked very much.
"A team of haymakers were assigned the task of scything
two meadows, one twice the size of the othor. Half a day
Fig. 6
1;3
Y3
Y6
!i3 Y3
Fig. 7
half the team can cut 1/3 of tho meadow in half a day. This
means that in the smaller meadow there is an uncut portion
equal to 1/2 - 1/3 = 1/6. If one scytheman can work 1/6
of a meadow in one day, and a total of 6/6 + 2/6 � 8/6
was cut, then there must be 8 workers.
"Tolstoy, who all his life enjoyed tricky problems that
were not too involved, learned about this problem from my
father when still a young man. When J met Tolstoy-al
ready an old man-and discussed the problem with him, he
was most delighted by the fact that the problem becomes
still clearer, literally transparent, if a very simple drawing
is employed in the solution (Fig. 7)."
!" - O B 9 1 49
We will now look [nio several problems that are mor�
easily solved arithmetically than algebraically if one has
quick wits.
Cows in the U
i eadow
PROBLEM
"'¥hen studying the sciences, problems are more useful
than rules," wrote J . Newton in his Universal Arithmetic
and accompanied his theoretical propositions with a number
Fig. 8
,. 51
and one cow will eat
1 + 24y
24·7 0 .
1 + 24y 1
24 ·70 1600 •
20 20.
and we have x =
Thus,
days.
cows would have eaten up all the grass in 96
52
Newton's P.'ohlelll
Now let us consider Newton 's problem of oxen, after
which the preceding one was patterned.
Incidentally, the problem was not devised by Newton
himself but is the product of folklore in mathematics.
"Three meadows covered with grass of the same thickness
an d rate of growth have the following areas: 3 ; hectares ,
24
4
10 hectares and hectares. The first served to feed 12 oxen
during weeks and the second, 21 oxen during 9 weeks.
How many oxen can feed on the third meadow in the course
of 18 weeksl"
SOLUTION
We introduce an auxiliary unknown y to denote what por
1
tion of the original supply of grass increases in growth over
hectare in one week. In tho first meadow, ono week sees an
incroase - of 3 � Y, and in four weeks the grass increases by
3 ; y ·4 ;0 y of the original supply over one hectare. This
=
(3 � + � y)
hectares. In other words, the oxen consumed as much grass
3
as covers a meadow with an area of ; + � y hectares. In
ono week, 12
oxen consumed a fourth part of this amount,
and ono ox in one week consumed a 4� th part, which is the
available supply over an area of
10+40y
144
hectares.
53
In exactly the same way, we fmd the area that foeds one
ox during one week from tho data for tho second meadow:
one-week growth over 1 hectare � y ,
nine-week growth over 1 hectare = 9y,
nine-week growth over 10 hectares � gOy.
The area of pasture land containing a supply of grass
for feeding 21 oxen during 9 weeks is
10 + 90y.
The area sufficient to feed one ox during one week is
1 0 + 90y 1 0 + 90y
9 · 21 189
54
Intel'cllanging the Hands of a Clock
PROBLEM
The biographer and friend of the eminent physicist Al
bert Einstein, A. Moszkowski, wished to distract his friend
during an illness and suggested the following problem
(Fig. 9).
The problem he posed was this: "Take the position of the
hands of a clock at 12 noon. If the hour hand and the minute
Fig. 9
{�-io=m
y x
5 - 00 �n
56
where m and n are integers (whole numbers) that can vary
between 0 and 11. From this system we fmd
'60 (!2m + n )
X= 143
60 (12n+m)
y
143
By assigning m and n the values from 0 to 1 1 , we can
determine all the required positions of the hands. Since
each of the 1 2 values of m can be correlated with each of the
12 values of n, it would appear that the total number of
solutions is equal to 12 · 12 = 144. Actually, however, it is
143 because when m = 0 and n = 0 and also when m = 1 1
and n = 1 1 we obtain the same position of the hands.
For m = 1 1 , n = 11 we have
x = 60, y = 60
and the clock shows 12, as in the case of m = 0, n =0.
We will not discuss all positions , but only two.
First example:
ni = 1, n = 1;
60 · 13 ;)
y=o-
5
x = 143 = 5 !T '
•
11
5
and the clock reads 1 hour 5 IT minutes; the hands have
•
x 143
�
143
The respective times are: 8 hours 28.53 minutes and 5
hours 42.38 minutes.
57
We know the total number of solutions: 143. To find all
the points of the dial that yield the required positions of
the hands, one has to divide the .circular dial into 143 equal
parts, thus obtaining the 143 desired points. At intermediate
points no other suitable positions can be found.
.58
Guessing Numbers
The reader is probably familiar with the game of guessing
numbers. The conjurer usually suggests performing opera
tions like the following: think up a number,i add 2, mul
tiply by 3, subtract 5, subtract the original number and so
on-a total of five, even ten operations. He then asks what
number you have and, with that answer, he gives you the
original number almost at once.
Tho secret of this "trick" is of course very simple, and
again equations give the answer.
Suppose the conjurer suggests a series of operations as
indicated in the left-hand column of the following table:
Think up a number
I x
add 2
I x+2
subtract 5
I 3x + 1
multiply by 2
I 4x + 2
subtract 1
I 4x+ 1
He then asks you for the final result and gives the answer
at once, How does he do this?
It is quite clear from a glance at the right-hand column of
the table, where the instructions of the conjurer have been
translated into the language of algebra. From thi.'; column it
is e"ident that if you think up a number x , then after all
the operations you get 4x + 1 . With this knowledge it is
easy to "guess" the original number.
59
Suppose, for example, you got the number 33. What the
magician does is solve the equation 4x + 1 � 33 fmding
x = 8. In other words, he takes the fmal result, subtracts
one (33-1 = 32) and then divides that number by 4 to
get 32 : 4 = 8. So the original number was 8 . If your final
result is 25, the magician does some mental arithmetic
(25 - 1 = 24, 24 : 4 = 6) and comes up with the number
you thought up, which was 6.
So you see it is very simple. The magician knows before
hand what has to be done with the result in order to obtain
the original number.
If that's so, then you can make your friends marvel by
letting them suggest the type of operations to be performed
on the thought-up number. You suggest that somebody
think up a number and perform operations of the following
nature in (lny order: add or subtract a known number (say.
add 2, subtract 5, and so on), multiply (but not divide
because that will greatly complicate the trick) by a known
number (by 2 OT 3 and so on), and then add or subtract the
original number. To get you completely confused, your
friend mounts operation upon operation. Say, he thinks
up 5 (which of course is kept secret), and then performs the
operations. It goes like this:
"I have thought up a number, multiplied it by 2, added 3,
and then added the original number; then I add 1, multiply
by 2, subtract the original number, subtract 3, and again
subtract the original number, and then subtract 2. Finally,
J multiply the result by 2 and add 3."
N ow convinced that he has you thoroughly muddled, he
gives the fmal number: " The result is 49."
To his great surprise, you nro the answer at him: 5.
HawaII that is done is now clear enough. When YOllr friend
states the . operations he is performing with the original
Dumber, YOH do the same with x. '¥hen he says " I have
thought up a number" you say (lo yourself): " I have x . "
He says, " I 've multiplied it by 2" (and he actually does
60
multiply the original number by 2), whereas you multiply
your x by 2 getting 2x. He adds 3 and you follow with 2x + 3,
and so forth. When at last he has you in a quandary with
his involved operations, you get wha t is shown in the follow
ing table (the left-hand column is what your friend says
aloud, and the right-hand column contains the operations
that you perform mentally:
I multiply by 2
I
I
x
2x
I
add 3 to tho result
I 2., + 3
add "1 ,
- I 3x -1- 4
milltijJly by 2
I ox 1 8
--
subtract 3
I 5.< +- ;)
I add 2 to it
I x+2
I now add 3
I 2.< + 7
I
'
subtract the original number x+7
add 5
I x+ 1 2
{ Xy
second one above, we see that
xy 2 =
4,
or 4 = 2. There are no numbers that satisfy this system.
�
5-0891 65
II. A different kind 01 surprise awaits us if we alter the
condition of the preceding problem somewhat. We assume
that the tens digit is 3 (not four) less that the units digit,
otherwise the problem remains the same. What number is
that?
Set up a n equation. If the tens digit is denoted by x,
then the units digit becomes x + 3. Expressing the rest of
the problem in the language of algebra, we get
10 (x + 3) + x - [lOx + (x + 3)1 = 27.
Simplifying, we arrive at the equality
27 = 27.
This is undoubtedly true, but it tells us nothing about the
value of x. Does this mean that there are no numbers that
satisfy the requirements of the problem?
Quite the contrary, it means that the equation we set up
is an identity: it is true for all values of the unknown x.
Indeed, it is easy to see that in this problem every two-digit
number in which the units digit exceeds the tens digit b y 3
has this property:
14 + 27 = 41, 47 + 27 = 74,
25 -I- 27 = 52, 58 + 27 = 85,
36 + 27 = 63, 69 + 27 = 96.
Ill. Find a three-digit number with the following prop-
erties:
(1) the tens digit is 7,
(2) the hundreds digit is less by 4 than the units digit,
(3) if the digits of the number are reversed, the new num-
ber will exceed the original one by 396.
Let us set up an equation and denote the units digit by x:
100x + 70 + x - 4 - [ 100 (x - 4) + 70 + xl = 396.
66
Aiter a few simp i iiications, th is equation yiei ds
396 = 396.
The reader already knows what this means: that every
three-digit number in which the frrst digit is 4 less than the
third (the tens digit plays no role) is increased by 396 if
the order of the digits is reversed.
Up to now we have been considering problems that are
more or loss artificial, bookish; their purpose was to help
acquire skill in setting up equations. Now that we are theo
retically equipped, let us tackle some problems of a practi
cal nature, from industry, everyday life, the military
sphere, and sports.
At the Hal'ber's
PROBLEM
Does algebra ever come in handy in cutting hair? Yes it
does. I became convinced of this when a barber once ap
proached me with an unusual request:
"Could you please help us here in a problem we have?"
"We've spoiled so much of the solution because of it!" put
in another harber.
"\'\That is the problem?" I asked.
"We have two solutions of hydrogen peroxide: a 30 per
cent solution and a3 per cent solution. We want to mix them
and get a 1 2 per cent solution, and we can't find the proper
proportion. "
I took a piece of paper and quickly figured out the proper
proportion.
It turned out to be very simple. What was it?
S O L UTION
68
N ow if tho tra m is coming towards me, then it will meet
me 4 minutes after the preceding one, and during the romain
, in/( (x - 4) minutes it will covel' the same distance that I
do in 4. minutes. This means that the distance I cover in one
minute is covered by the tram in X�4 minutes.
We get the equation
12-x :r - 4
---n- = - 4-
----+ 5 h
A B
I
A B
!
I
)?
i
A
r- _____________ -f I
._ 1
Fig. 10
SOLUTION
We denote by x the time (in hours) it takes the boat to
cover the distance between A and B in standing water (at its
70
own speed), and by y the time it takes the rafts to cover that
distance. Then the boat does 1/x of the AB distance in one
hour, and the rafts (going with the current) do 11y of that
distance. Thereforo when the steamship goes downstream it
covers 1/x + 1 1y of the AB distance, and going upstream
(against the current) it does 1/x - 1 1y. Now it is stated in
the hypothesis of the problem that going downstream the
ship does 1/5 of the distance in one hour, and going upstream,
{++f = +'
117. This gives us the following system of equations:
1 1 1
-X - y = T '
Note that in solving this system it is best not to get rid of
the denominators, simply subtract the second equation from
the fIrst. This yields
2 2
y 35
and so we get y = 35. The rafts cover the distance from A
to B in 35 hours.
Two Cans of Coffee
Pl\OBLEM
Two cans containing coffee have the sarno shape and a1'O
made out of the same tin. One can weighs 2 kg and is 12 em
high; the other weighs 1 kg and is 9.5 em high. Find the net
weigh t of the coffee in both cans.
SOLUTION
Denote the weight of the larger can by x , and that of the
smaller One b y y. Let the weights of the cans themselves be z
+ z
y + t = l.
2,=
and t, respectivel y. We then have the equations
{X
71
Since the weights of the contents of full cans are in tim
same ratio as their lJolumes, that is! as the cubes of their
heights, * it follows that
x
9 "
=-
123
- or x = 2.02 y.
y . ....
"" 2.02
Now the weights of the empty cans are in the same ratio
as their total surface areas, which is to say I as the squares of
their heights. Therefore
z
9 " "" 1 . 60
-=-
122
or z= 1 . 60t.
t ."
Thus, the net weight of the coffee in the larger can is 1.92 kg,
and that in the smaller can is 0.94 kg.
A. Question of Ha.udlln
PROBLEM
At a party, 20 people danced. Mary danced with seven
partners, Olga with eight, Vera with nine, and so forth up to
* This proportion is applicable only when the till of the cans i�
thin. This i s hec;:ll1se, strictly spe;1kin � , the ollter and inner surfacos
of cans are lIot similar; what is more, the height of the inside part of
a can is, strictly spealdng, different from the height of the can itseH.
72
Nina who danced with all the partners. How many men
partners were there at the party?
S O LUTION
Reconnaissanee at Sea
P[10IlLEM No. 1
74
from which we get
1
x=2Z'
Fig. 11
they meet every 10 seconds; when they go in the same direc
tion, one catches up with the other every 170 seconds. Find
the speed of each cyclist if the circular track is 170 metres
long.
75
S O L UTION
A Competition of Motorcyclists
PROBLEM
7(\
SOLUTION
Although we are seeking seven unknowns, we will solve
the problem by seeking two: we will set up a sy�tem of two
aqua tions in two unknowns.
Denote the speed of the second motorcycle by x. Thon the
speed of the frrst one is x + 15 and of the third x - 3.
The total path length we denote by y. Then the time en
route comes out to
y
x + 15 for the fil'st motorcycle,
x
.!!.. for the second,
"
x-3 for the third.
We know that the second motorcycle took 12 minutes (or
1/5 hour) more to cover the distance than the flrst. And so
1
-; - x + 15 = 5 '
Y Y
77
Hemoving brackets and collecting terms, we get
3x - 225 = 0
and ftnally
x = 75.
Knowing x, we can find y from tho nrst equation:
JL _ JL _ �
5
=
75 90
or 90.
-
y
So now we have the speeds of the three motorcycles
90, 75, and 72 km per hour.
And the path length is 90 km.
Dividing the path length by the speed of each motorcycle,
we can find tho time en route:
1 hour for the first motorcyclo,
1 h and 1 2 minutes for the second,
1 h and 15 minutes for the third .
We have thus found all the seven unknowns.
AVel'age Speeds
lPIlOBLEM
An automobile covers the distance between two cities at
a speed of 60 km per hour, on the return route the driver
does 40 km an hour. Find the average speed.
SOLUTION
78
This "simple" solution would be correct if the trip there
and back lasted the same time. But it is clear that the return
trip (at a smaller speed) must have taken a longer time than
the trip thore. Taking that into consideration, we can see
that 50 is not the answer.
True enough, an equation yields a different answer. It is
easy enough to set up an equation if we introduce an auxi
liary unknown, namely the quantity I for the distance be
tween the cities. Denoting the sought-for average speed by
x, we get the following equation:
21 1 1
-;- = 60 + 40 '
Since I is not equal to zero, we can divide through by
to get
whence
x
2
= 1 = 48 .
60+40
1
�= i.. + i..b
x a
6-0891 81
eiem"nts by "-" signs. Let us examine the upper storage
location (a dot indicates where the fractional part of A num
ber begins and a dashed line separates the first digit, which
records the sign of the uumber, from the other digits).
The recorded number reads, in binary, +1011.01 or, in the
ordinary decimal system, 1 1 .25.
Memory locations afO also used to record instructions
(commtmds) that make IIp the program . Let us see what kind
,
,
,
1+1-1 I i-I 1+1-1+1+1 1+1-1 HI
, , ,
.oPeration I I : II I III
j ! I 1-1-1+1+1 H I 1+1+1 t I 1+1+1+1 I 1+1 1+/+1
I i
, ,
Fig. 12
6· 83
2nd instruction: multiply the number in the 4th 10catioll
b y itself (it is squared) and put the result, or 1' , on a card
(the arrow indicates a printout of the final result).
3rd instruction: transfer of control to the 1st location.
In other words, the instruction "transfer of control" means
that all instructions must again be carried out in order,
beginning with the first. So we again have the 1st instruc
tion.
1st instruction: add the numbers in the 4th and 5th loca
tions, and again record tho result in the 4th location. Thus
in the 4th location we have the number 1 + 1 = 2:
(4) 2,
(5) 1.
2nd instruction: square the number in the 4th location
and write out the result, 01' 22 , on a card (the arrow indi
cates a printout).
Brd instruction: transfer of control to the first location
(which means a transfer to the 1st instruction again).
1st instruction: send the number 2 + 1 = 3 to the 4th
location:
(4) 3,
(5) 1.
2nd instruction: print out the number 32,
3rd instruction: transfer of control to the 1st location and
so on.
We see that the computer computes the squares of the
inteKers and prints them out on a card. Note that it is not
necessary to write in by hand the next new number because
the machine goes through the sequence of integers and
squares each of them. Following that program, the computer
finds the squares of all whole numbers, say, from 1 to
10 000 in the course of a few seconds or even fractions of
a second.
84
I t must be noted bowever that actually the program
for computing the squares of integers must be somewhat
more complicated than that described above. I n parti
cular this refers to the second instruction. The point is that
printing the result on a card takes"much more time than com
puting a single operation. For this reason, the results are
stored in free storage locations of the memory unit and
are then later printed out on cards "at leisure", so to say.
Thus the flrst flnal result is stored in the 1st free location,
the second result in the 2nd location, the third in the 3rd
location, and so on. This was omitted in the description
given above.
Another thing. The computer cannot be engaged for a long
time computing squares because there are not enough loca
tions in the memory unit; also thero is no way of "guessing"
when the machine has computed enough squares to be able
to turn it off on time (remember the computer does many
thousands of operations per second). To handle this situa"
tion, special instructions are included in the program to stop
the computer at the right time. For example, the program
may state that the computer is to work out the squares of
all numbers from 1 to 1 0 000 and then stop.
There are of course more sophisticated instructions that
we will not go into here so as not to complicate our discus
sion.
Here i s a real program for fmding the squares of all inte"
gers from 1 to 1 0 000:
Program la
(1) addition 8 9 8
(2) multiplication 8 8 10
(3) addition 2 6 2
(4) conditional jump 8 7 1
(5) stop
(6) 0 0 1
(7) 10 000
85
(8) 0
(9) 1
(10) 0
(11) 0
(12) 0
88
with numerical values of the coefficients a, b, c, d, e, I ,
a', b', . . . .
Here is all appropriate program :
Program 2
(1) X 28 30 20 (14) + 3 19 3 ( 26) a
( 2) X 27 31 21 (15) + 4 194 (27) b
(3) X 26 30 22 (16) + 5 195 (28) c
(4) X 27 29 23 (17 ) + 6 196 (29) d
(5) X 26 31 24 (18)jump 1 (30) e
(6) X 28 29 25 (19) 6 6 0 (31) t
(7) - 20 21 20 ( 20) 0 (32) a'
(8) - 22 23 21 (21) 0 (33) b'
(9) - 24 25 22 (22) 0 (34) c'
(10) : 20 21 -+ (23) 0 (35) d'
( 1 1 ) : 22 21 -+ (24) 0 (36) e'
(12) + 1 19 1 (25) 0 (37) f'
(13) + 2 19 2 (38) a"
86
Seventh instruction: from the number in the 20th location
subtract the number in the 21st location, and again send
the result (ce - bf) to the 20th loca tion.
Instructions eight and nine are carried Ollt similarly.
Thus, in the locations 20,21, 22 we have the following num
bers:
(20) ce - bt
(21) ae - bd
(22) at - cd
Instructions 10 and 11: the quotients
ce-bf a/-cd
ae bd and ae bd
are formed and are printed out on a card (which means they
are given as a final result). These are the values of the
unknowns obtained from the first system of equations.
Tbe first system bas thus been solved. Why are any further
instructions needed? The next portion of tbe program (loca
tions 12 to 19) is needed to prepare the computer for solving
the second system of equations. Let us see how this is dOlle.
Instructions 10 to 17 consist in the following: to the con
tents of locations 1 to 6 we add the material in location 19,
and the results again remain in locations 1 to G. Thus, after
the 17th instruction has been carried out, the first six loca
tions will look like this:
(1) X 34 36 20
(2) X 33 37 21
(3) X 32 36 22
(4) X 33 35 23
(5) X 32 37 24
(6) X 34 35 25
Instruction 18: transfer of control to location ODO.
In what way do the new notations in the first six locations
differ from the earlier material? In that the first two address-
90
es in these locations have the numbers il:': to 37 instead
of 26 to 31. This means that the computer will again perform
the same operations, hut this time it will take numbers from
locations 32 to 37 (instead of from 26 to 31) where the coef
ficients of the second system of equations are located. The
computer thus solves the second system of equations, and
then goes On to the third system, and so forth.
From what wo have seen it is clear that the most important
thing is to write a proper program. By itself the computer
is quite helpless and can't do anything. All it can do is car
ry out a program of instructions. There are programs for
computing roots, logarithms, sines, for solving equations
of high degrees and many more. There are even programs
for playing chess; as wo have SOOll, and for translating from
one language into another, albeit rather poorly. A computer
can do a lot of things, and of course the more complicated
the assignment, the more involved is the program of instruc
tions.
We conclude with a word about so-called compiling rou
tines, which are programs by means of which the computer
itself can work out a program for solvi�g some problem.
This greatly simplifies tho writing o[ programs, which can
often be an extremely time-consuming matter.
Chapter three
AS AN AID '1'0 AIU'I'HM)<;'1'IC
�nstantaneous Multiplieation
.
Calculating prodigies (sometimes called lightning cal
culators) often simplify their computational work by resort
ing to simple algebraic manipulations. For example,'to
square H88 one does as follows:
988 988 = (988 + 12) . (988 - 12) + 12'
= 1000·976 + 144 = 976 144
It is easy to see that the calculator here made use of the
familiar algebraic rule
� = � - � + � = � + � � - b) + b�
We can make good use of this rule in oral?:calculations.
For example,
27' = (27 + 3) (27 - 3) + 3; = 729,
63' = 66 ·60 + 3' =�3969,
92
18' = 20 · 16 + 2' = 324,
37' = 40 ·34 + 3' = 1369,
48' = 50 . 46 + 22 = 2304,
54' = 58 · 50 + 4' = 2916.
= 1000 (986-3) + 14 · 3
The last line depicts the device that the calculator uses.
Here is a nice way of multiplying two three-digit numbers
in which the number of tons is tho sarno, whereas the sum
of the units digits comes to 10. For example, to multiply
783 · 787
do as follows:
78 · 79 = 6162, 3 · 7 = 21
to get
616 221 .
93
The reasoning behind this procedure becomes clear from
the following manipulations:
(780 + 3) (780 + 7)
� 780 · 780 + 780 · 3 + 780 · 7 + 3 · 7
� 780· 780 + 780 · 1 0 + 3 · 7
� 780 (780 + 10) + 3 · 7 780 , 790 + 21
�
95
The foregoing permits us to assert that
3862567 ends in 6,
81572:1 ends in 5,
491 1 7 3 2 ends in 1, and so forth.
Infinite "Numbers"
There are also longer strings of digits that appear at the
end of certain numbers and again appear in their product.
We will now show that there are an infmity of such strings
of numbers.
96
We know two-digit strings (groups) of digits that have
this property: 25 and 76 . To n.ud three-digit groups, adjoin
in front of 25 or 76 a digit such that the resulting three
digit group of digits has the required property.
What digit should we adjoin to the number 76? Let us
denote it by k. Then the desired throe-digit numLer is
100k + 76.
The general expression for numbers ending in this group of
digits is:
1000a - i- 100k + 76, 1000b + 100k + 76 and so on.
Let us multiply together two numbers of this type. We get
1 000 OOOab + 100 OOOak + 100 OOObk + 76 OOOa
+ 76 OOOb + 10 OOOk' + 15 200k + .5 776.
All terms except the last two end in at least three zeros.
For thi s reason the product ends in 100k + 76 if the differ
ence
15 200k + 5 776 - (1 00k + 76) = 15 1 00k + 5700
= 15 OOOk + 5000 + 100(k + 7)
7-0891 97
end in 10001 + 376? If we remove the,brackets in this prod
uct and discard all terms endi ng in four zeros and more,
then we get the terms
752 0001 + 141 376.
The prod,.ct ends in 10001 + 376 if the difference
752 0001 + 141 376 - (10001 + 376)
= 7.5 1 0001 + 'l41 000
>11 Note that the two-digit group 76 may be found via reasoni ng
similar to that given above: all one needs to do is fin d the digi t to be
adjoined on the left to the digi t 6 so that the resulting twcrdi git group
has the desired propert y. Therefore, the "n umber" .,_ 7 109 376 can be
obtained by adjoining d i� its to 6 on the left on c aftcr another.
** Infin ite "n umbers' may also be considered in systems of n ume
rati on other than the deci mal system. Numbers considered in a sys
tem of numerati on wi th base p are termed p-adie numbers .
,. 99
Allditional Payment
A PROBLEM OF F O L K L O H E
Divisibility by 11
where the dots stand for the sum of higher orders. From
N we subtract the number l1(b + 10e + toOd + . . . ),
which is a multiple of tt. Then the resulting difference,
which is readily seen to be equal to
a - b - to (e + tOd + . . . ),
t01
will have the same remainder upon d ivision b y 1 1 as the
number N. Adding to this difference the number 11 (c +
+ 10d + . . . ), wbich is divisible by 1 1 , we get the number
a - b + c + 10(d + . . . ) ,
whicb also has the same remainder upon division by 11 as
N. From it we subtract tbe number 11(d + . . . ), wbich is
divisible by 11, and so on. We finally get
a - b + c - d + . . . = (a + c + . . . ) - (b + d + . . . ).
This number has the same remainder after dividing b y
1 1 as does the original number N.
From this we get the following criterion for divisibility
by 1 1 : from the sum of all digits in odd sites, subtract tbe
sum of all digits standing in even sites; if the difference
is 0 or a number (positive or negative) divisible b y 1 1 ,
then the number being tested i s a multiple of 1 1 ; otherwise,
our number is not exactly divisible by 1 1 .
Let u s test the number 8 7 635 064:
8 + 6 + 5 + 6 = 25,
7 + 3 + 0 + 4 = 14,
25 - 14 = 11.
Thus, the given number is d ivisible by 11.
There is another procedure for determining divisibility
by 11 which is convenient for numbers that are not very
long. It consists in the number under test being split from
right to left into groups (or blocks) of two digits each and
then the groups being added. If the resulting sum is exactly
divisible by 1 1 , then the original number is a multiple
of 11, otherwise it is not. Suppose we want,to test the number
528. Didding it into the appropriate groups of two (5/28)
and adding the groups, we get
5 + 28 = 33.
102
Since 33 is exactly divisible by 1 1, so also is the number
528:
528 : 1 1 = 48.
We now prove tbis criterion of divisibility. Split a multi
digit number N into groups of digits. We obtain two
digit (or one-digit *J numbers which we designate from
right to left as a, b, c, and so on I so that the number N can
be written as
N = a + 100b + 10 OOOe + . . . = a + 100(b + 100e + .. . ).
Subtract from N the number gg (b + 100e + . . . ) , which.
is divisible by 1 1 . The resulting number
a + (b + 100e + . . . ) = a + b + 100 (e + . . . )
will have the same remainder, when divided by 1 1 , as the
number N. From this number we. subtract 99 (c + . . . ) ,
which is divisible by 11, and so on. Finally, we find that the
number N has the same remainder upon division b y " 1 1 as
the number
A License Number
PROBLEM
Three students of mathematics out for a stroll noticed
a car break the traffic regulations. Not one of the students
noticed the number on the license plate (it was a four-digit
number). but, being mathematicians, they noticed some
peculiarities about the number. One recalled tbat the first
two digits were the same. A second recalled that the last
two digits were also alike. And finally the third student
* If the number N had an odd number of di gits, the last (leftmost)
group will be a onc-digit block. Besides, a block of the form 03 sh o uld
also be regarded as a one-digit number 3 .
103
maintained that the four-digit number was a perfect square.
Is this information enough to find out the number on the
license plate?
SOLUTION
Denote the first (and second) digit of the desired number
by a, the third (and fourth) by b. Then we have the number
1000a + 100a + lOb H � 1100a + 1 1 b � 11 (100a + b).
This number is divisible by 1 1 and so, being a perfect square,
is also divisible by 11'. Wbich means the number 100a +
+ b is divisible by 1 1 . Using either one of the two earlier
devised criteria for divisibility by 11, we find that 1 1 di
vides the number a + b. And this means that
a + b � 11
since each of the digits a, b is less than ten.
The last digit b of the number, which is a perfect square,
can assumo only the values
0, 1, 4, 5, 6 , 9.
And so for the digit a, which is equal to 1 1 - b, we get
the following possible values:
1 1 , 10, 7, 6, 5, 2.
The flfSt two are unsuitable and that leaves us the following
possibilities:
b = 4, a = 7;
b � 5 , a � 6;
b � 6 , a � 5;
b � 9, a � 2 .
We see that the license number can ollly be one of the fol
lOWing four:
7744, 6655, 55G6, 2299.
104
But the last three of these numbers are not perfect squares:
the number 6655 is divisible by 5 but not by 25; tbe number
5566 is divisible by 2 bu t not by 4; the number 2299 =
= 121 ·19 is not a square either. That leaves only one num
ber, 7744 = 88', which is the solution.
ntvlsibility by 19
4 704588 1 1
+2·
47045 1 UO
+ 18
47 06 1 3
+6
471 1 2
+4
47 1 5
+ 10
5T7
+ 14
19.
Composite Numbe.'s
There are an infinity of the so-called primes (prime
numbers) , which are integers exceeding unity that canDot
be divided b y any whole number other than unity and the
integer itself.
The sequence of primes begins 2, 3, 5, 7, 1 1 , 13, 17, 19,
23, 29, 31, . . . , and, as the dots indicate, extends indefini
tely. These primes interpose tllemselves in the range of
composite numbers,. and split the natural-number sequence
into more or less extended portions of composite numbers.
How long are these portions? For example, can we find
a sequence of, say, a thousand composite numbers with not
a single prime among them?
It can be proved, although this may seem improbable,
that straight strings of composite numbers between primes
may be of any imaginable length. There is no bound to
the length of such strings: they may consist of a thousan d ,
a million, a trillion and so on composite numbers.
For the sake of convenience, we will make use of the
symbol n!, which stands for the product of all numbers from
1 to n inclusive. For instance, 51 = 1 · 2 · 3 ·4 ·5. We will
+1
107
now prove that the sequence
[(n + i)l + 2 1 , [(n + i)l + 31, [(n + i)l + 4] ,
up to [(n + i)l + n + il inclusive
consists of n successive composite numbers.
These numbers follow one another in the sequence of the
natural numbers since each succeeding onO is 1 greater
than the preceding number. It now remains to prove that
they are composite.
The first number
(n + 1) I + 2 = 1 ·2 ·3 ·4 · 5 ·6 ·7 · . . . . (n + i) + 2
is evon since both terms contain the factor 2. Now, every
eVOn number greater than 2 is a composite number.
The second number
(n + 1) 1 + 3 = 1 . 2 · 3 ·4 · 5 · . . . . (n + 1) + 3
consists of two terms, each of which is a multiple of 3.
Hence, this number too is composite.
The third number
(n + 1) 1 + 4 = 1 ·2 ·3 ·4 ·5 · . . . . (n + 1) + 4
is exactly divisible b y 4 because it consists of terms that
are multiples of 4.
In similar fashion. we find that the number
(n + 1)1 + 5
is a multiple of 5. and so forth. In other words. each number
of our sequence contains a factor that is different from unity
and from itself; which means it is a composite number.
If you want to write, say, five composite numbers in
succession, all you need to do is substitute 5 for n in the
sequence given above. Yon will then get the following se
quence:
722, 723. 724, 725, 726.
108
This is not the only sequence made up of five composite
numbers. There are other sequences, such as
62, 63, 64, 65, 66.
Or take the smaller numbers
24, 25, 26, 27, 28.
Let us now try to solve the following problem.
Vvrite down a sequence of ten composite numbers.
SOLUTION
109
and add unity. This gives us
NI + 1.
Since this is an integer I it must contain at least onO prime
factor. which means it must be divisible by at least one
prime. But. by hypothesis. the sequence of primes does not
exceed N. yet the number NI + 1 cannot be exactly divid
ed by any number less than or equal to N. for every time
we obtain a remainder of 1 .
Thus, w e cannot assume that the sequence of primes is
finite: the assumption leads to a contradiction, and this
means that no matter how extended the string of composite
numbers in the sequence of natural numbers, we may rest
assured that at the end of the sequence we will encounter an
infinity of primes.
1+ 90 000 000000
1+
000
We now carry out the division of the numerator by the
denominator:
1 80 000 000 000 I gO 000 000 001
90 000 000 001 1.999 999 099 977 . . .
899 999 999 990
810 000 000 009
899 999 990 810
8 1 0 000 000 009
899 ggg g9S 010
810 000 000 009
899 ggg 980 01 0
810 000 000009
899 999 800 010
810 000 000 009
899 998 OOU 01 U
810 000 000 009
899 980 OUU 01 U
81 0 000 000 009
899 800 UOU UW
810 000 000 009
898 UUO OUU 010
810 000 000 009
88U UUO OUU 01 0
8 1 0 000 000 009
700 UOO UOU 01U
630 000 000 007
70 UOO 000 003
H2
This is clearly a monotonous, gruelling task where errors
can crop up at any stage. Yet it is important when solving
this problem to note the exact point at which the sequence
of nines is broken and a different sequence of digits sets in.
N ow notice how easily algebra handles the situation.
It makes use of the following approximate equality: if
a is an extremely small fraction, then
1
� 1 -a
I+a
--
8-0891 113
It shows that due to the smallness of the velocities under
consideration as compared with the velocity of light, it is
hardly possible to detect any departure from the old law of
compo�ition of velocities. Even with such groat velocities as
one kilometre per second, a difference shows up only in the
11th digit of the number being determined (in ordinary
engineering calculations, one confines himself to 4 to 6
digits). We Can therefore state very definitely that the new
Einsteinian mechanics practically changes nothing in engi
neering calculations that deal with the "slow" (compared
with the velocity of light) bodies. There is, however, a field
of modern life where this conclusion calls for caution. It
is the field of space flight. Today artificial satellites and
space vehicles have reached velocities of the order of
10 km/s. Here the discrepancy between classical and Ein
steinian mechanics appears in the ninth digit. And there
are higher velocities in the oiling . . . .
114
by 5 yield a remainder of 4
by 6 yield a remainder 01 5
by 7 yield a remainder oX 6
by 8 yield a romainder of 7
by 9 yield a remainder of 8
SOLU TION
Buyiug a Sweatm'
PROBLEM
You have to pay 19 rubles for a sweater. All you have
on you are three-ruble bills and the cashier only has five
ruble bills. How can you pay for the sweater? Or maybe
you can't?
The question comes down to this: how many three-ruble
bills do you have to give to the cashier to pay 19 rubles
and receive change from the cashier in the form of five
ruble bills. There are two unknowns in this problem: the
number x of three-ruble bills and the number y of five
ruble bills. But there is only one equation that can be set up;
3x - 5y = 19.
Although one equation in two unknowns has an infinity
of solutions, it is not at all obvious that there is even one
among them that has integral positive values of x and
y (recall that these are the numbers of bills). That is why
algebra has worked out a method for solving such indetermi
nate equations. The credit for introducing them into algebra
belongs to the first European representative of that science,
the eminent mathematician of antiquity Diophantus, whence
the term D iophantine equations.
SOLUTiON
We will use the above example to show how such indeter
minate equations are solved.
We have to find the values of x and y in the equation
3x - 5y = 19
knowing that x and y are positive integers (whole numbers).
116
First we isolate the unknown with the smallest coeffi
cient, the term 3x, and get
3x � 19 + 5y
whence
1 9 + 5y 6 1 + 2y
x� 3
+y+ 3 '
6
Since x, and y are integers, the equation can only hold
true if 3- is a whole number as well. Let us denote it
l-I-2y
-
by t. Then
where
and, hence,
3/ � 1 + 2y, 2y � 3t - 1.
From the latter equation we can determine y:
3t-1 !oil t-1
2- '
2 = /+ -
y � --
Since y and / are integers, it follows that '-;-1 must like-
wise be some whole number /,_ Consequently,
y � / + I,
and
'- 1
tl � -2-
whence
2/, = t - 1 and t � 2/, + 1.
Put the value / = 2t, + 1 into the preceding equations:
y � / + /, � (2/, + 1 ) + /, = 3/, + 1 ,
x = 6 + y + / � 6 + (31, + 1 ) + (2/, + 1) = 8 + 51,.
117
And so for x and y we have found the expressions •
x = 8 + 5tv
Y = 1 + 3t,.
Now we know that the numbers x and y are not only
integers but are also positive, or greater than 0. Consequently,
8 + 5t, > 0,
1 + 3t, > 0.
From these inequalities we find
5t, > - 8 and
3t, > - 1 and
Such are the restrictions on t,: it is greater than -1/3
(and, hence, all the more so greater than -8/5). But since
t, is a whole number, we conclude that it can be only one
of the following values:
� = 0, 1 , 2, 3, 4, , . . .
1 18
or you pay 1 3 three-ruble bills and receive 4 five-ruble
bills:
13.3 - 4·5 � 19,
and so on.
Theoretically, the problem has an infinity of solutions,
but in practical situations the number of solutions is limit
ed because neither the buyer nor the cashier has an infinity
of bills. For instance, if each has 10 bills, the payment can
be made in only one way: by paying 8 three-ruble bills and
getting 5 rubles in return. We have thus seen that ind eter
minate equations are capable of yielding quite d efinite
pairs of solutions in practical problems.
Returning to our problem, we suggest that the reader,
as an exercise, work out a variant in which the buyer has
only five-ruble bills and the cashier only three-ruble bills.
We then get the following series of solutions:
x � 5, 8, 1 1 ,
Y � 2, 7 , 12, . . . .
Indeed,
5·5 - 2.3 � 19,
8·5 - 7·3 � 19,
11 · 0 - 1 2 . 3 � 19.
119
we find (knowing that x < 0 and y < 0)
8 + 511 < 0, 1 + 311 < 0
and, hence,
8
1' < - 5'
Auditing Accounts
PROBLEM
901"
fi'ie:ze
Jl'mdf'?.5 tJ/
at .f9r. 36'£. a m?tf'/!
120
The number of metres sold was blotted out, but obviously
it was not in the form of a fraction; the sum of money re
ceived was also partially blotted with only the last three
digits distinguishable, and it also was evident that there
were three preceding figures.
Was it possible for the auditors to recover the original
figures from those distinguishable in the entry?
SOLUTION
121
is a whole number and since 2 cannot ·be divided by 125,
.
It follows that 17-4x
125 must be a whole number, which we
denote by I.
Then from the equation
t
17 --4x
----rzs =
we get
17 - 4x = 1251,
where
1 -'
/' = -4-
and, hence,
411 = 1 - I,
1 = 1 - 411,
X = 1251, - 27,
y = 61711 - 134".
We know that
100 <; y < toDD.
Consequently,
100 <; 6171, - 134 < toDD
and from this we have
122
It is clear that t, can have only one integral value:
I, = 1
and then
x = 98, Y = 483,
Buying Stamps
PROBLEM
x+ y + z = 40,
where x is the number of one-kopeck stamps, y is the number
of 4-kopeck stamps, and z the number of 12-kopeck stamps.
Subtracting the second equation from the first, we get
one equation in two unknowns:
3y + 11z = 60.
We find y to be
y = 20 - 1 1 . ; .
Clearly, ' is a whole number. Denote it by t. Then we
;r
have
y = 20 - f 1 t,
z = 3t.
123
Substituting the expressions for y and z into the second
of the original equations, we get
x + 20 - 11t + 3t = 40
and
x = 20 + St.
t=
1 0 1 1
Check
X= 1 20 1 28 20 · 1 + 20 · 4 + 0 · 1 2 = 100,
28 . 1 + 9 · 4+ 3 . 12=100.
Y= 1 20 1 9
Z= 10 1 3
Buying Fruit
PROBLEM
Guessing a Birthday
PROBLEM
126
1 - I = 51" I = 1 - 51"
Y = 2(1 - 51,) - 2t, = 2 - 12t"
x = 1 4 - 3(2 - 12 t,) + 1 - 51, = 9 + 31t,.
Knowing that 31 ;;. x > ° and 12 ;;. y > 0, we find the
bounds of t,:
9 I
-- -
31 < t 1 < 9
Consequen tly,
I, = 0, x = 9, Y = 2.
The birthday falls on the 9th day of the second month,
which is February 9.
Another solution that dispenses with equations can also
be tried. We are told the number a = 12x + 31y. Since
1 2x + 24y is divisible by 12, the numbers 7y and a have
the same remainders when divided by 12, Multiplying by
7, we find that 49y and 7a have the same remainders upon
d ivision by 12. But 49y = 48y + y and 48y is divisible
by 12. This means y and 7a have the same remainders upon
division by 12. In other words, if a is not divisible by 12,
then y is equal to the remainder upon division of the number
7a by 12; but if a is divisible by 12, then y = 12. This
gives the number of the month y quite definitely. But if
we know y then it is easy enough to find x.
A tiny piece of advice: before finding the remainder
obtained from the division of 7 a by 12, replace the number
a by its remainder obtained from division by 12. That's
much simpler. For example, if a = 170, then do the fol
lowing mental arithmetic:
170 = 12 ·14 + 2 (and so the remainder is 2),
2 ·7 = 14; 14 = 12 ·1 + 2 (hence y = 2),
170-31y 170-31 · 2 108
X= 12 = 12 12 = 9 (and so x = 9) .
You can name the birthday of your friend: it is February 9.
127
Now let us prove that this trick never fails, in other
words, the equation always has only one solution in positive
integers. Denote the number given to you by your friend
by a so that finding the birthday reduces to solving the
equation
12.x + 31 y = a.
We assume the contrary. Suppose that the equation has
two distinct solutions in positive integers, namely the
solution Xl' Yl and the solution xz, Y z (Xl and Xz do not ex coed
31, and Yl and y, do not exceed 12). We then have
1 2.xl + 31Yl = a,
12x2 + 31Y2 = a.
Subtracting the second equation from the first, we get
12(Xl - x,) + 31( Yl - y,) = o.
From this equation it follows that the number 1 2(xl -
- x,) is divisible by 31. Since Xl and x, are positive numbers
that do not exceed 31, their difference Xl - x2 is less than
31. Therefore the number 12(Xl - x,) will be': divisible
by 31 only when Xl = X" that is, when the first solution
coincides with the second. Thus, the assumption that there
are two distinct solutions results in a contradiction.
Selling Chickens
AN OLD PROBLEM
Forenoon x y , m
{
A few manipulations yield
(m - n)x + 1 0n � 35,
(m - n) y + 16n � 35,
(m - n)z + 26n � 35.
Subtracting the flrst equation from the third and then
the second from the third , \ve get, in succession,
( (m - n) ( z - x) + 16n = 0,
l (m - n) (z - y) + 10n = 0,
$I-OB91 129
{ (m - n) (x - z) � 1Bn,
(m - n) (y - z) � 10n.
Now divide the nrst of these equations by the second:
x-z 8
s
x- z
�
or
y -z
-- = -;:- -
130
Now, going back to the equations
m:!' + n( 10 - x) = 35,
my + n( 16 - y) = 35,
mz + n(26 - z) = 35
and substituting into tbem the values of x, y and z thus
found, we obtain the prices at which the chickens were sold:
3 1
m=3T rubles, n = 1 T rubles.
Thus, in the forenoon the chickens sold for 3 rubles and
75 kopecks, and in the aftemoon for 1 ruble and 25 kopecks.
132
whence
2y
x= y -2 '
133
R emoving brackeLs and simplifying, wo got
xz = yt,
where x, y, Z, t are integers less than 10. To nnd the solu
tions we set up pairs of equal products made up of 9 digits:
1 ·4 = 2·2 2·8 = 4·4
1 ·6 = 2·3 2·9 = 3·6
1·8 = 2·4 3·8 = 4·6
1·9 = 3 · 3 4·9 = 6 ·6
2 · 6 = 3·4
Thero are nine equalities. From each ono it is possible
to set up one or two desired groups of numbers. For example,
using the equality 1 · 4 = 2· 2 we fmd one solution:
12 ·42 = 2 1 · 24.
TTsing 1 · 6 = 2 . 3 we get two solutions:
12·63 = 21 · 36, 13 · 62 = 31 · 26.
In this manner we obtain tbe following 14 solutions:
12 ·42 = 21 · 24 23.96 = :)2 . 69
1 2 · 63 = 21 · 36 24 · 63 = 42·36
1 2 · 84 =
21 · 48 24 · 84 = 42 . 48
13 · 62 = 31 . 26 26 · 93 = 62 . 39
1 3 · 93 = 31 ·39 34 · 86 = 43 · 68
14 . 82 = 41 · 28 3G · 34 = 63 · 48
23 · 64 = 32 ·46 46 · 96 = 64· 69
Pythagorean Numbers
A convenient and very exact method used by surveyors to
lay down perpendicular lines consists in the following.
Suppose it is required to draw a perpendicular to the straight
line MN through point A (Fig. 13) . To do this, from A lay
B a a a A N
M
a
a
a a
a
a
a
a
Fig. 13
135
They are called Pythagorean numbers. According to the
Pythagorean theorem, such numbers may serve as the sides of
a rigbt triangle, and so a and b are termed tbe legs and c
is the hypotenu,e.
Clearly, if a, b, c is a Pytbagorean triad (a triplet of
Pythagorean numbers), thenpa, pb, pc, where p is an integral
factor, are Pythagorean numbers too. Conversel y, i f any
Pythagorean numbers have a common multiple, then that
multiple can be used to divide through all the numbers and
again obtain a Pythagorean triad. Thereforo, to begin with
let us investigate only triplets of relatively prime Pythago
rean numbers (the others can be obtained from them by mul
tiplication by an integral factor p).
We will now show that i n each of these triplets a, b, c
one of the legs must be even and the other odd. We assume
the contrary. If both legs a and b are even, then the number
a' + b' will be even and hence so also will the hypotenW!e.
However, this contradicts the fact that the numbers a, b, c
do not have any factors in common since throe even numbers
have the common factor 2. Thus at least one of the legs,
a or b, must be odd.
There is still another possibility: both legs are odd and
the hypotenuse is even. It is easy to show that this cannot
be. J ndeed, if the legs are of the form
2x + 1 and 2y + 1 ,
C - b = n2.
137
Solving this system, we get
m2 + n2 m2 _ n2
I 2-"--
c= 2 b = --;;
for m = 13, n = 5
for m = 9 , n = 7 63' + 16' = 65'
for m = 11 , n = 7 77' + 36' = 85'
138
All other Pythagorean triads either have common factors or
contain numbers exceeding a hundred.
Pythagorean numbers have a number of curious properties
that we list below without proof:
( 1 ) One of the legs must be a multiple of three.
(2) One of the legs must be a multiple of four.
(3) One of the Pythagorean numbers must be a multiple
of five.
A glance at the Pythagorean numbers given above will
convince the reader that these properties do hold true.
An Indeterminate Eqnation
of the 'l'hh'd Degree
The sum of the cubes of three integers may be a cube
of a fourth number. For instance, 33 + 43 + .53 � 63•
Incidentally, this means that a cube whose edge is equal
to 6 em is equal in size to the sum of three cubes whose
Fig. 14
140
If we know two quadruples of numbers that satisiy the orig
inal equation, we can find the new qnadruple: to do this, add
to the numbers of the first quadruple the numbers of the
second quadruple multiplied by k, where k has the value
given above.
To make use of this procedure , it is necessary to know
two quadruples of numbers satisfying the original equation.
We already know one: (3, 4, 5, -6) . Where can we flll d
another one? This is very simple. For the second quadruple,
take the numbers r, -r, s, -s, which obviously satisfy
the original equation. In other words, set
c � 5, d � -6,
a = r, � = -r, y � s 6 � -so
144
theorem in the margin of a book by Diophantus (as he did
a numher of other theorems in the field of number theory)
and added these words : HI have discovered a truly marvel
ous demonstration, which this margin is too narrow to
contain."
Nowhere has this proof been found, either in the papers
of the great mathematician or in his correspondence or
anywhere else.
Fermat's successors were forced to work out the matter in
their own way.
Here are the results of these effort" Euler (1797) proved
Fermat's theorem for the third and fourth powers; the futh
power was proved! by Legendre (1823), the seventh' b y
Lame and Lebesgue (1840). In 1849 Kummer proved the
theorem for a broad rango of powers and, incidentally, for
all exponents less than one hundred. These latter investiga
tions go far beyond the limits of the realm of mathematics
known to Fermat and it is quite a mystery how Fermat
could have di,covered a general proof of hi' great theorem.
True, he could have been mistaken.
For those interested in the history and the present state
of Fermat's problem we suggest A. Ya. Khinchin's Fermat's
Great Theorem (in Ru.sian), a nice booklet that can be read
by anyone with an elementary knowledge of mathematics .
10-0891
Chapter five
Besides the notation ·Via we can also make use of an, which
is particularly convenient in the sense of generalization,
for it demonstrates very pictorially that every root is nothing
other than a power whose exponent is a fraction. It was
proposed by the well-known Flemish engineer and mathe
matician of the 16th century Simon Stevin.
Which Is G.'eater?
PROBLEM 1
147
SOLUTION
148
Subtract 253 from each and then compare them:
27 and 20 Y57.
Since y 57 is groater than 2, it follows that 20 V 57 > 40;
hence
V 3 + Y 19 > V 7 + Y 10.
Solve It at a Glance
PH OBLEM
Take a close look at the equation
XX' =3
and find x immediately.
SOLUTION
Anyone familiar with algebraic symbols will ftgure out that
3.
3/
x=v
Then
x = ,;
3/ y.
149
and the equation becomes
('yy)V = 3
Of, cubing,
yV 3' .
=
It is clear that =
y 3 and, hence,
x = v3/-y = v3/.3.
Algebraic Comedies
P H O B LEM 1
The sixth mathematical operation makes it possible to
devise actual algebraic comedies and farces on such topics
as 2 · 2 = 5, 2 = 3 and the like. The humour in these mathe
matical shows lies in the fact that the error-a rather ele
mentary one-is somewhat camouflaged and is not at once
apparent. Let us take two plays from this comic repertoire
from the field of algebra.
To start with,
2 = 3.
We begin with the unquestional equality
4 - 10 = 9 - 15.
Then to both sides of the eq uality we add the same quan-
1
tity, 6"4 :
(2-{)' = ( .3-{)'
follows
2 -. 25 = 3 - 25 ,
From the fact that the squares are equal it does not at all
follow that the first powers are equal. Say, (-5)' = 5',
but -5 does not equal 5, Squares may be equal even when
the first powers have different signs , That precisely is the
case in Ollr problem:
( _ �)2= (�)".
But -� is not the same as + '
PROBLEM 2
Here's another algebraic farce (Fig, 1 �):
2 , 2 = 5,
It follows tbe patb of the preceding problem and is based on
the same trick We start out with the undoubtedly flawless
151
16-36 = 2'5-45
16 -36720)4 =
= 2 '5- 4 5 + 20).4
(4_�)2 (5-tY
=
2x2=5
Fig. 15
equality
16 - 36 = 25 - 45.
Equal numbers are added to each side:
1
16 - 36 + 20 4 = 25 - 4S + 20 41
and then the following manipulations are carried out:
4' - 2 · 4 · { + ( f) 2 = 5' _ 2 . 5 . ; + ( ;) 2 ,
( 4 - f)2 = (S-f) 2.
152
Again using the false conclusion of the earlier problem,
we finally get
9 9
4- 2= 5 - 2 ,
4 = 5,
2 · 2 = 5.
Those amusing instances should bo a warning to the
inexperienced mathematician in performing imprudent opor·
ations with equations involving tho radical sign.
Chapter SIX
ShakiB" Hands
PROBLEM
A meeting gathered Hnd someone counted the total number
of handshakes to be 66. How many people were there at the
meeting?
SOLUTION
Algebraically, the problem is solved with great ease.
Each of x people shook the hand of x 1 persons, which puts
-
Y= V �.
Then x = 2y2 and we obtain
y+ 1�' + 2 = 2y2 or 2y2 - 9y - 18 = O.
1 55
Solving this equation. we get two values for y;
0
y , = 6. y, = - 2 '
-\i- + g8 - 72 + 2 = 6 + 64 + 2 = 72.
,V ; '2
A Troop of l\lonkeys
PROBLEM
SOLUTIO1\'
,7
1)
Solving this equation, we obtain
I, = 1 and t, = 4.
The ball will be at a height of 20 metres twice: after
a lapse of 1 second and after 4 seconds.
This appears to be improbable and so without giving
much thought to the matter we discard the second solution.
But that is a mistakel The second solution is quite meaning·
ful, for the ball did indeed reach a height of 20 metres twice.
First when it went up, and a second time on the way down.
It can easily be figured out that with an initial velocity
of 25 metres a second the ball will spend 2.5 seconds on
its upward leg reaching a height of 31.25 metres. After 1
second it will reach 20 metres, but will.. go on upwards
)nother 1.5 seconds. Then it will take the same amount of
time to drop back to the 20·metres level and, a second
later, will reach the ground.
EuleJ" s l'roblem
Stendhal, in his autobiography, relates the foll owing
about his days of schooling:
"The mathematics teacher had a book by Euler and there I
found his problem on the number of eggs that a peasant
woman was carrying to market . . . . This was a revelation
to me. I realized what it meant to use the tool called algebra.
But, the devil take it, nobody had ever told me about this
thing . . . . "
Here is the problem from Euler's Introduction to A lgebra
that so strongly impressed the young Stendhal.
Two peasant women together took 100 eggs to market, one
had more than the other. Both sold them for the same sum
of money. The first then said to the second: "If I had had
your eggs, I would have earned 1 5 kreuzers," to which the
158
second repiied: If 1 had had your eggs, I would have earned
6 i- kreuzers." How many eggs did each have to begin with?
SOLUTION
Suppose the first peasant had x eggs and the second 100 - x.
If the first had had 100 - x eggs, she would have earned 15
kreuzers. This means the first woman sold her eggs at
15
100- x
apiece.
In the same way we find that the second peasant sold.her
eggs at
2 20
6 "3 : x = &;"
apiece.
We can now determine the actual earnings of each peasant
woman:
15 15x
first: x · 100 - x 100 �-'x 1
20
second: (IOO -x) . &;" = 20 (100-x)
3x
•
Simplifying we get
x' + 160x - 8000 = 0
and from this,
X, = 40, x, = - 200.
Here the negative root is meaningless, and the problem
has only one solution: the first peasant woman brought
40 eggs to market and the second, consequently, 60.
159
This problem can be solved faster, but i t requires a good
deal of insight and is harder to hit upon.
Suppose the second peasant ha d " times the number of eggs
of the first. Since they earned the same money, the first
peasant sold her eggs at a price " times that of the second.
If they had exchanged their goods before selling them, the
first peasant would have k times as many eggs as the second
and would have sold them at k times the price. Which mea.ns
she would have earned k' more money than the second
peasant. And so we have:
k' = 15 : 6 ; = ;; = :
whence
Loudspeakers
PROBLEM
Fig. 16
falls off with the square of the distance, we have the eqlla�ion
1) = (50-x)'
4 x2
1 1- 0 8 9 1 161
which is taken as the positive direction when we set up the
equation.
If we lay off 1 00 metres from the location of the four
loudspeakers in the required direction, we find the point
where the sound from both groups of loudspeakers comes in
with equal intensity. This point lies at a distance of 100 me
tres + 50 metres � 150 metres from the group of nine
loudspeakers.
To summarizo, thon, we have found two points of equal
audibility (that is, from among the points lying on a straight
line joining the sources of sound). There are no other such
points on that line, but outside the line there are. It can
be proved that the set of all points satisfying our problem
constitutes a circle drawn through the two points that were
found as through the endpoints of the diameter. It will
be seen that this circle bounds a rather extensive area
(cross-hatched in the drawing) inside which the audibility
of the group of four lou dspeakers is greater than that of the
group of nine, and outside this circle the situation is just
the opposite.
384000km
Fig_ 17
A Hard Problem
Tn 1895 N. Bogdanov-Belsky painted a picture called
A Hard Problem (see Fig. 18) and many of those who have
seen it most l ikely skipped over the problem itself, though
it is well worth looking into. The idea behind the picture is a
problem in mental arithmetic, to be solved at a glance:
1 02 + 1 1 2 + 122+13 2+ 142 -?
365 -.
The problem really isn't easy. But the pupils of the teacher
portrayed in the pjcture-and the portrait is an accurate one
of S. A. Rachinsky, professor of natural science, who left the
university to become an ordinary schoolteacher in the vil
Jage-coped with it. In his school this talented teacher
cultivated habits of mental arithmetic based on a marvellous
handling of the properties of numbers. The numbers 10, 1 1 ,
166
Fig. 1 8
Fiudiun
- -
NlImbeJ's
-
P R O B LEM
Two Tl'ains
PHOBLEM
Two railway lines intersect at right angles. Two trains
arc racing at tho sarno timo to the intersection, one having
left a station 40 km from tbe intersection, the other from
a station 50 km from the intersection. The first train is
doing 800 metres per minute, the second, 600 metres per
minute.
In how many minutes after start will the locomotives be
separated by the shortest distance? Find that distance.
SOLUTION
Let us make a diagram of the movements of the trains.
Let the straight lines AB and CD be the intersecting lines
(Fig. 19). Station B is 40 km from the point of intersection 0,
170
station D is 50 km from it. Sup pose that after a lapse of
x minutes the locomotives are separated by the smallest
distance MN = m. The train that left B would by then
have covered the distance BM = 0.8x, since it does 800 mo
tres, or 0.8 km, a minute. And so OM = 40 - 0.8x. In the
A A
M
50 76
9.6
0 � N D
C N [) C 12.8
M
40
B B
fig. 19 f'ig. 20
The minus sign means that the locomotive has passed the
intersection by 9.6 km. And the distance ON is equal to
50 - 62 ·0.6 = 12.8
which means the second locomotive is 12.8 km short of the
intersection. This p osition of the locomotives is shown in
Fig. 20. As we now see, it doesn't look at all like what
we imagined before we began. The equation turned out to be
very tolerant and despite the incorrect drawing gave us the
proper answer. This tolerance is clearly due to the algebraic
rules for signs.
Fig. 21
SOLUTION
Denote by a the distance AD (from A to tIlt' foot of the
perpendicular BD to AD) and by x tho distanc" C D . Then
A C � AD CD = a
- -x and CB = YCD ' �I· BD'
= Yx + 20'. The time during which a train covers the
'
distance A C is
AC a-x
0.8 = --0:8 '
The travelling time by highway from C to B is
GB
- y ,' + 20'
0.2 0.2
The time required t o got from A to B is
a-x Vx2. + 202
0.8
_
+- �
0 . 2 :::...
...c..::
The equation
a-x -V�
--0:8 + 0.2 =m
173
can be cast in the form]
.
-
x -V� a
0.8 -I- 0.2 = m - 0.8 .
Multiplying b y 0.8, we get
-x + 4 V x' + 20' � 0.8 m - a.
Denoting 0 . 8 m - a by k and getting rid of the radical
� ign, we obtain the quadratic equation
15x' - 2kx + 6400 - k' 0, �
whence
k+ V' I M2 -go 000
x= 15
Since k = O.8m - a, it follows that Ii. attains a mInImum
when m is a minimum and conversely. * But for x to be reaC
16k' must be a t least equal to 96 000. Hence that figure
is the smallest we can have for 16k'. Therefore, m becomes
least when
16k' 96 000 =
and, consequently ,
k+0 Y60()()
x � 15 = -
1-
5
- :::; 5.16.
To summarize, then, no matter how long a = AD is,
the flag station must be located at about 5 km from the
point D .
Quite naturally, our solution is meaningful only for
cases where x < a, because when we set up the equation
we regarded the expression a - x as being positive.
If x = a :::; 5.16, then there is no need to build a flag
station at all and the highway will have to be built straight
* Note that k > 0 since
O.8m = a - x + 4 -Vx2 + 202 > a -- x + x= a.
174
to the main station. And the same goes for cases where the
distance a is shorter than 5.16 km.
This time, it turns out, we have more insight than the
equation. If we were to blindly follow the equation, we would
have to construct a flag station right after the station, which
would he nonsense: in that case x > a and therefore the
time
a-x
o.s
during which we travel by rail is nega tive. This is an instruc
tive example that shows caution is necessary when handl ing
rna thcma tical tools and interpreting resul t8. One must bear
in mind that the final figures may be meaningless if the
promises on which tho usc of those tools is based arc not
properly taken into account.
An Optimal Hi ghway
PROBLEM
A town A is located on a river and we have to send freight
to n, a town located a kilometres downstream and d kilo
metres "rom the river (Fig. 22). The problem is to locate
Fig. 22
175
a highway between B and the river so that transportation
of goods from A to n is cheapest, taking into consideration
that the transport cosL of a ton-kilometre on the river
is half that by highwa y.
SOLUTIO:\'
176
Now, an angle w]l Ose sine is equal to �;;
is 1300: This mean s
the highway must be sited at an angle of 60° to tbe river,
no matter what the distance AC.
Here again we have the same peculiarity encountered in
tilO preceding problem. The solution is meaningfnl only
lInder a certain condition. If tbe point is located so that tbe
h ighway built at an angle of 60° to the river passes on the
otber side of the town A , then the solution cannot be applied;
in that case the bighway should be built directly between D
and the town A , dispensing with river transport.
12-0891 177
And the product will be greatest when x 0 , which is the
a
=
The numbers x and y are positive. The third' part will obvi
ously be
a
3 + Y - x.
p times q times
'-v-' ��---
px q
-'-'-:----'- = x + a - x = a ,
_ + (a-x)
q p
that is, the quantity is a constant.
On the basis of what has been proved above (pp. 177·178)
we conclude that the product
x x x a-x a-x a-x
q q
-
q
- 0 0 -
p p p
reaches a maximum when all its separate factors are equal,
that is, when
x a-x
p q
Knowing that a x =
- y, we obtain (after a simple mani
pUlation) the proportion
x p
-
y q '
180
Thus, the product xPy' attains a maximum (given the
constant sum x + y) when
x ; y = p ; q.
In the same fashio l we can prove that the products
xPyqzr, xPyqzTtU and so forth
(given the constant sums x � y + z, x + y + z + t and
so on) attain maximum values whenever
x : y : z = p : q : r, x:y:z : l = p:q:r:u and so on.
A Beam of U
l aximulII Volume
PROBLEM
The problem is to saw out of a cylindrical log a rectangular
beam of largest volume. Find the shape of the cross section
it will have (Fig. 23).
181
}'ig. 23
SOLUTION
If the sides of tho rectangular cross section aro x and y,
then by the Pytllagorean theorem we have
' '
x + y � d'
where d is the diameter of the log. The volume of the beam
is a maximum when the area of its cross section is a maxi
mum, that is, when xy becomes a maximum. Now if xy
is a maximum, then so is the product x2y2 . Since the sum
2 2
x + y is constant, it follows by what has been already
proved that the product x'y' is the largest possible one when
x2 = y2 or x = y.
Hence the cross section of the beam must be a square.
ltIakill!J a Kite
PROBLEM
We have a kite in the shape of a circular sector and it is
required to change the shape so that it will have the largest
possible area for the given perimeter. What will tho shape of
the sector be?
SOLUTION
183
The quantity S reaches a maximum for the same value of x a s
the product 2x (l - 2x) , which is four times the area. Since
the sum of the factors 2x + (I 2x) = I is a constant quan
-
t'ig. 24
Suppose x metres of the old wall are retained and the re
maining 1 2 - x metres are dismantled so that the materials
obtained are used again in the construction of a part of the
wall of the new house (Fig. 25). If the cost of laying a lineor
12-'
Fig. 25
185
metre of wall using new material is equal to a) then repair
ing x metres of the old wall will cost ;: ; erecting a portion
a (12 -x)
of length 12 - x will cost 2 ; the cost of the remain
ing portion of that wall will be a [y (12 - x)] or a (y +
-
Suppose the portion along the high fence (seo Fig. 26) is
x, while the width (that is, the dimension of the lot perpen
dicular to tho high wooden fenco) is equal to y. Thon x + 2y
Fig. 26
metros of fencing is needed to enclose that portion, so that
x + 2y = I.
Tho area of the lot is oqual to
S = xy y (I - 2y).
=
187
It altains a maximum value at the same time that the quan
tity
2y (l - 2y)
tloes. This is twice the area and is a product of two factors
with a constant sum I. And so to obtain the greatest area we
must have
2y = 1 - 2y
whence
I I
y -= 4 ' x = I - 2y = z ·
In other words, x = 2y, or the length of the plot must be
twice its width.
I
t'ig. 28
Fig. 27 Fig. 29
188
ways, as shown in Fig. 28. Of what width must the lateral
strips be and at wbat angle are they to be bent for tho cross
section of the trough to have the maximum area (Fig. 29)?
SOLUTION
Fig. 30
The quantity
y, z, as 3S2, and the latter Can be exprossed as a product:
(y -I- z) (y -I- z) (x -I- z) (3x - 3z).
The sum of these four factors,
y -I- z -I- y -I- z -I- x -I- z -I- 3x - 3z = 2 y -I- 4x = 21,
189
is a constant. Therefore the product 01 our lour factors is a
maximum when they are equal to each other, or
y + z � x + z and x + z � 3x - 3z .
From the first equation we have
y = X,
H � V -- r2 � VrR2
R2 - - _
x'
4n'
_
do.
j91
Since
( -,,-
2n
- ) 2 + R2 - ( -"--
2n
) ' = RZ
is a constant, it follows (on the basis of what was proved on
pp. 179-181) that the last product is a maximum for the
value of x when
( 2il
X
) [. R2_ ( 2il
2
:
X 'J
) = 2: 1
and from this
( 2: ) 2 = 2R2 _ 2 ( 2Xn ) 2,
3 ( : r 2R'
" = and x= f2; R V 6 "" 5. 15R.
In degrees, the arc x "" 2950 and, hence, the arc of the
sector that was cut out must contain �65°.
� )
�
M e B N
Fig. 32
192
SOLUTION
It might seem that the lower the flame the better. This h
not so: in a low position of the flame of the candle , the rays
fall at a very sloping angle. But if the candle is high and
the rays fall at a sharp angle, then the light source is too
far away. The best illumination is clearly at some interme
diate height of the flame above the table. We denote it by x
(Fig. 32). And we use a to denote the distance BC of the
coin B from the foot C of a perpendicular passing through
the flame A . If the brightness of the flame is i, then the
illumination of the coin is given, by tho laws of optics, as
i t cos a
COS " =
AS' ("V a' + x'l'
where " is the angle of incidence of a pencil of rays AB.
Since
x
cos ex = cos A = -=-
AS = -Va2 .� x2.
it follows that the illumination is
�x2)3 - � )
(a2 (X2 (/Z)2 ( 1 - x2��a2
)2 (
1-
a'
= ( x2 +a2 ).
1
x Z + a2
\3-089\ 193
The transformed expression roaches a maximum together
with the expression
'" 195
Using the conditions given in the problem, we set up the
following two equations:
{X + tx + y) + (x + 2y) -+ (x -I- 3y) -+ (x -+ 4y) = 100,
7 Ix + (x + y)J = (x + 2y) + (x + 3y) + (x + 4y).
B D
I
----+- - ± - - - -
I
-- +
I I
j
I
G
t
-
I.
- -- -- -- -- -- g
-- -
_ _
Iii. .! .f
2
-- - - t
-- - I -- :
----
. }I �
I
..
£
.J ..L 3
-- - I
--
I
-' ------1
- � ..l
4 I ;:
r-
I
.l. __
I
�
c - - II
5
_
_
�
I
A
I--- -�
- I I I
Fig. 33
two equal figures: ABDC and DGEC. The area of each de
scribes the sum of the terms of our progression. Hence, the
double sum of the progression is equal to the area of the
rectangle ABGE, or
(AC + CE) ·AB.
But AC + CE gives the sum of the flfSt and fifth terms of
the progression; A B is the number of terms in the progres
sion. Therefore, the double sum:
28 = (the sum of the extreme tenns) . (the number of terms)
or
s _ (first term +-- last term) .(number of terms)
- .
2
197
Watel'ing the Garden
PROBLEM
Thore are 30 rows in a garden, each row 16 metres in length
and 2.5 metres in width. The gardener waters his garden by
hauling pails of water from a well 14 metres from the edge
of the garden, and then walks between the rows. One trip
to the well su1nces to water only one row.
What distance does the gardener cover in watering the
whole garden? The start and nnish are at the well.
SOLUTION
To water the flfst row, the gardener covers the d istance
14 + 16 + 2.5 + 16 + 2.5 -+ 14 � 65 metres.
In watering tho second row, he covers
14 + 2.5 -I- 16 + 2.5 + 16 + 2.5 + 2.5 + 14 � 65 +
+ 5 � 70 metres.
Each subsequent row requires 5 metros more to be covered
than the preceding one. We then get the progression
65, 70, 75; . . . ; 65 + 5 ·29.
The sum of the terms of the progression equals
(65+65 ·�29·5) 30
2
4125 metres.
In all, the gardener covers a distance of 4.125 km in
watering his garden.
199
whence
x � 3ly � 496.
The supply that was put in came to 496 decalitres of feed
calculated \0 last 16 weeks.
A '1'eam of lUggers
PROBLEM
200
The number y cannot be zero and so we can cancel it out
of the equation to get
6x = 24
and
x = 4.
This means the last one of the tea m to start working was
busy 4 hours.
We have found the answer to the problem, but if we had
been curious enough to ask how many participated in the
work, we wouldn't have been able to say, despite the fact
that that number entered into the equation (it was y). It is
merely that there is not enough information in the hypothe
sis of the problem to get that answer.
Apples
PROBLEM
A man has an apple orchard and sells to his first customer
half of all the apples plus half an apple; to tho second cus
tomer he sells half of the rest plus half an apple; to the
third, half of the remainder plus half an apple, and so on.
To the seventh customer he sells half of what remains and
another half-apple. And that is all he had. How many apples
did the farmer start out with?
SOLUTION
Use x to denote the original number of apples; then the
first customer received
x+ 1 x+1
-2-
_.
2 2-
the second customer,
1. ( _ x + l ) + 1. = x + l
2 x 2 2 P '
201
the third,
x+1 x + + -.!. = I) x+1
(
2 x 2 4 2 23 1
_ _
-.!.
and the seventh customer bought
x+1
---v- .
We thus have the following equation:
x+l x+1 + x+1 + + x+1 =
- 2- + � 23 . . . 2" x
or
(1 1 1 )
(x + 1) 2+ 2' + y, + . . . + 2' = x.
1
Buying a norse
PROBLEM
In the old Russian arithmetic of Magnitsky we fmd an
amusing problem that I give here in a translation into
modern language.
Somebody sold a horse for 156 rubles. However the buyer,
after all, decided not to take the horse and so returned it
to its owner with the words:
"Thero is no point in my taking this horse for such a price
because the animal is just not worth it."
202
Then the owner suggested the following.
"If the price of my horse is too high, then buy only the
horseshoe nails, and then I'll give you the horse free of
charge. There are six nails in each horseshoe. For the first
nail you give me only + of a kopeck, for the second � , for
the third 1 kopeck, and so on."
Fig. 34
204
or
�ls of Logarithms
Before the invention of logarithms, the requirements of
speedy calculations gave rise to tables of a different kind
in which the operation of multiplication is replaced by sub
traction instead of addition. These tables are based on the
identity
(a+b)' (a- b)'
ab -
-
4 4
207
All one needs to do is open the brackets to see that the iden
tity holds.
Having ready at hand the fourths of squares, one call fmd
the product of two numbers without performing the multipli
cation but simply by subtracting the fourth of the square of
the difference of the numbers from that of their sum.
The same tables simplify squaring numbers and taking the
square roots of numbers, and when combined with a table of
reciprocals, they simplify the operation of division as well.
Their advantage over logarithmic tables is that they yield
exact, not approximate, results. On the other hand, however,
they are inferior to logarithmic tables i n a number of ways
which are of more practical importance. Whereas tables of
the fourths of squares permit multiplying only two numbers,
logarithms enable one to find at once the product of any num
ber of factors, and , what is morc, to raise a number to any
power and extract foots having arbitrary indices (integral
and fractional). For example, it is impossible to compute
compound interest with the aid of tables of fourths of
squares.
Even so, tables of fourths of squares continued to be
published after the appearance of a great variety of loga
rithmic tables. In 1856, a set of tables appeared in France
entitled:
A table of the squares of numbers from 1 to 1000 million
with the aid of which one can find the exact product of numbers
by an extremely Simple procedure that is more convenient than
by means of logarithms. Compiled by A lexander Cassar.
This very idea pops up time and again without their
inventors realizing that it is an old one. I personally was
approached b y two inventors of similar tables and they were
greatly surprised to learn that their invention was already
over three hundred years old.
A different and younger rival of logarithms are the COID
putation tables of most engineering reference works. These
are combination tables with columns of squares, cubes, square
208
roots ! cube roots, reciprocals! circumferences and areas of
circles for numbers from 2 to 1000. Such tables are very
convenient for many engineering calculations but they are
not always suffICient. Logarithmic tables have a far broader
range of application.
Logarithmic-'1'able Champions
Whereas the computational needs of practical life and
general engineering are quite satisIied with three- and four
place tables, the tbeoretician has need for tables witb many
more places than even the fourteen-place logarithms of
Briggs. Generally speaking, in most cases a logarithm is an
irrational number and cannot be exactly expressed by any
number of digits; the logarithms of most numbers are given
only approximately, no matter how many decimal places
are taken-the larger the num.ber of decimal places in the
mantissa , the more exact the result. Evon the Briggsian four
teen-place tables* prove insufflCient for some scientifIC work.
However, the researcher will nOvor go unsatisfied, for among
the 500 various types of logarithmic tables that have ap
peared since logarithms were invented there will definitely be
one that can bandle the job. Let us take, for example, the
twenty-place logarithms of numbers from 2 to 1200 that
were published in France by Callet (1795). For smaller ranges
of numbers there are tables of logarithms with fantastic
numbers of decimal places. There are giants that many mathe
maticians have never even suspected existed.
Here is a short list of the champion logarithms (they are
all natural logarithms, not common logarithms) : "
*' The fourteen-place logarithms of Briggs are, incidentally, avail
able only for the numbers from 1 to 20 000 and from 90 000 to 101 000.
** Natural logarithms use the base 2.718 . . . (instead of 10). They
are discussed later on.
210
48-place tables of Wolfram for the numbers up to 10 000;
61-place tables of Sharp;
102-place tables of Parkhurst;
and, finally, the logarithmic wonder of them all: the
260-place logarithms of Adams.
The last case, to be exact, is not a table but only the so··
called na tural logarithms of five numbers: 2, 3 , 5, 7 and 10
and the 260-decimal conversion factor for converting them
to common logarithms. But it is easy to see that Witll the
logarithms of these fIve numbers it is possible, via simple
addition or multiplication, to obtain the logarithms of a
multitude of composite numbers; for example, the logarithm
of 12 is equal to the sum of the logarithms of 2, 2 and 3, and
so forth.
Another logarithmic marvel is the slide rule ("wooden
logarithmsll) but it has become so common a counting tool of
the engineer as to be on the level of yesteryear's abacus
among clerical workers. The slide rule is such a routine tool
we are no longer amazed that while it operates on the prin
ciple of logarithms, the user need not evOn know what a loga
rithm is.
1\umber
I Lo,;arithm
I Nllmber
I Log�rithm
2 0 .30 11 1 .04
3 0 .48 12 1 . 08
4 0 . 60 13 1 . 11
5 0 . 70 ! 14 1 . 15
15 1 . 18
I
6 0 . 78
7 0 .85 16 1 .20
8 0 .90 ! 17 1 .23
9
!I
U .95 18 1 .26
19 \ . 28
212
The mathematical trick that amazed you is this:
log '1'(:55 digits) � s.\3; . . .
The desired logarithm can lie between
� and 3�'190 or between 1 . 09 and 1 . 13.
In this interval we have the logarithm of only one whole
number, namely 1 . 1 1 ; it is the logarithm of 13. That is
exactly how the startling result was found. True, to do all
this mentally in a flash requires the training and skill and
wit of the professional, but essentially it is quite simple,
as you can see. Tricks of this kind are now within your grasp,
if not mentally then at least on paper.
Suppose you are asked to find the 64th root of a 20-digit
number.
Without even asking for the number you can state the
result of the extraction: the root is 2.
Y
Indeed, log ; (20 digits) � 1964, .
Consequently, i t lies
between � and 1�499 , or between 0.29 and 0 . 32 . There is
only one such logarithm for a whole number: 0.30 . . . , or the
logarithm of the number 2.
You can even demolish your questioner by telling him what
number he was about to propose: the famous chess number
2" � 18 446 744 073 709 .5.5 1 616.
x = 1:1 500 V - ;.
Using a table of logarithms, we fmd that
x = 10 300.
Which means the ox requires 10 :1 00 calories.
*' In contrast to the "productive" ration, which is the part that
goes to build up the animal for slaughter.
Logarithms ill Music
Musicians do not often take a liking to· mathematics; most
of them respect the science but prefer to stay away from it.
Yet even those musicians who do not verify "harmony by
means of algebra" (like Pushkin's Salieri) come into contact
with mathematics much more frequently than they even
suspect; what is more, their contact is with such frightful
things as logarithms.
I permit myself here a short quotation from an article b y
the later physicist Professor A . Eichenwald. I t appeared in
the Russian Astronomical Calendar for 1.91.9 and was entitled
"On Large and Small D istances".
"A friend of mine from Gymnasium days liked to play the
piano but detested mathematics. He spoke with a touch of
scorn about music and mathematics having nothing what
soover in common. 'Truo, Pythagoras found some kind of
relationships between sound vibrations, but it is precisely
the Pythagorean scale that turned out to be unsuitable for
our music. '
"I magine the surprise of my friend ' when J showed him
tha t in running his hands over the ke ys of a modern piano he
was actually playing on logarithms. Indeed, the steps of
the temperea chromatic scale are not arranged at equal distan
ces either with respect to the number of vibrations or with
respect to the wavelengths of the appropriate sounds; they
are the logarithms of these quantities. Only the base of the
logarithms is 2 instead of 10, as commonly used.
"Suppose the note do of the lowest octave (we will call
it the zero octave) is given as n vibrations per second. Then
the note do of the f,rst octave will have 2n vibrations, that
of the mth octavo, n ·2m vibrations, and so on. Let us denote
all notes of the chromatic scale of a piano by the numbers p,
assuming the fundamental tone do of each octave to be the
zero tone. Then, for example, sol is the 7th tone, la the 9th,
and so on. The 12th tone is again do, only an octave higher.
215
In a tempered chromatic scale, each tone has �·2 more vibra
tions than the preceding tone. That means that the number
of vibrations of an.y tone can be expressed by the formula
Npm= n · 2m (;YZ)p.
"Taking logarithms, we get
log 2
log Npm = log n + m log 2 + P 12
or
log N pm = log n -+ ( m . / . ; ) log 2.
2
log N pm = m + ;2 .
15-0891 217
loudness of sonnd expressed in beh is equai to the comrrton
logarithm of its physical intensity.
A few examples will help to clarify this matter.
The soft rustling of leaves is estimated at 1 bel, a loud
conversation is put at 6 . 5 bels, the growl of a lion at 8.7 bels.
From this it follows that the sound intensity of a conversa
tion exceeds the rustling of leaves by a factor of
10 6 . 5 - 1 = 10' 5 � 316 000.
The growling of a lion is louder than a conversation by a
factor of
10 8 . 7 - <;,5 10 2. 2 = 158.
=
2200
whence
log x = 12 (log 25 - log 22), x = 4.6.
A gas-fIlled lamp emits 4.6 times more light than a vacu
um lamp. Thus, if a va cuum lamp is rated at 50 watts, then
the gas-filled lamp will yield 230 watts under the same con
ditions.
Let us calculate further to find out what increase in abso
lute temperature (in per cent) is necessary to double the
brightness of the lamp.
SOLUTION
We set up the equation
(1 + X )"
tOO
= 2
and find tha t
log ( 1 + t�o ) = l� 2 and X= 0%.
we fmd that
x = 1 . 1 3.
The brightness increases 1 3 % .
Calculating for a two per cent increase in temperature,
we find a 27 per cent increase in brigbtness; if the tempera
ture increases 3 % , the brightness will increase 43% .
It is DOW clear why so much attention in the manufacture
of electric light bulbs is paid to increasing the temperature
of the filament, with every extra degree at a premium.
223
The Number e
( 1 + +r
as n increases without bound.
For many reasons which we cannot go into hero tho number
e is highly desirable as a base for logarithms. Such tables
(tables of "natural logarithms") exist and are extensively
used in science and engineering. Tho champion logarithms
involving 48, 61, 102 and 260 digits that we spoke of a little
while ago use the number e for their base.
The number e often puts in an appearance where it is least
of all expected. Let us take a look at the following problem.
How should one partition a given number�a so that the
product of all its parts is a maximum?
We already know that the largest product for a constant sum
is obtained when the numbers are all equal. Clearly, the
number a is to be partitioned into equal parts. But into how
tnany equal parts? Two, three or ten? Techniques in higher
* Also, this nnmber, Hke re, is transcendental, which means it
Mnnot be obtained by solving any algebraic equati An invl)lvin2' _intQ�
fitra l coefficient�,
224
mathematics enable us to establish that the largest product
is obtained when the parts are as close as possible to e.
For example, partition 10 into a number of equal parts
such that they are as close as possible to 2.718 . . . . To do
this we have to fmd the quotient
10
2 . 718 . .
3.678 '"
( In ' � 37,
(�) 5 = 32.
Tn order to obtain the largest product of the parts of 20,
the number has to be partitioned into 7 equal parts because
20 : 2.718 . . . � 7 . 36 "" 7.
The number 50 has to be partitioned into 18 parts and the
number 100 into 37 parts because
50 : 2.718 . . . � 18.4"
100 : 2.718 . . . � 36.8.
The number e plays a tremendous role in mathematics,
physics, astronomy and other sciences. Here are some of the
questions considered mathematically that involve e ( the
list could be extended indefinitely ) :
225
Barometric height formula (decreasing pressure with in
creasing height),
Euler's formula (see the second part of my Physics for
Entertainment (Mir Publishers, Moscow): the chapter entitled
uJules Verne's Strong Man and Euler's Formula"),
The law of cooling of bodies,
Badioactive decay and the age of the earth,
Oscillations of a pendulum in the air,
Tsiolkovsky's formula for rocket speeds (see my book
Tnterplanetary Travel [in HussianJ),
Oscillatory phenomena in a radio circuit,
The growth o/. cells,
A Logarithmic Comedy
.
PI\OBLEM
Here is another one of those mathemalical comedies played
out in Chapter 5: prove that 2 > 3. This time we make use
of logarithms. The comedy starts out with the inequality
4 >8'
! 1
= -log , log, V (V 2.
It is easy to see that this equation is true. In deed,
!
N = - log 2 10g, V V .
��
V V i,
N times
. .
MIR
encounter equations, logarithms,
roots, progressions, the ancient
and famous Djophantine analysis
and much more. The examples afe . Publishers
pictorial, vivid, often witty
and bring out the essence of the Moscow
matter at hand. There are